• Users Online: 965
  • Print this page
  • Email this page

 Table of Contents  
Year : 2023  |  Volume : 6  |  Issue : 1  |  Page : 1-3

Artificial intelligence in airway management and anaesthesia

Department of Anaesthesiology, Critical Care and Pain, Tata Memorial Hospital, Homi Bhabha National Institute, Mumbai, Maharashtra, India

Date of Submission02-Mar-2023
Date of Acceptance03-Mar-2023
Date of Web Publication20-Apr-2023

Correspondence Address:
Prof. Sohan Lal Solanki
Department of Anaesthesiology, Critical Care and Pain, Second Floor, Main Building, Tata Memorial Hospital, Parel, Mumbai - 400 012, Maharashtra
Login to access the Email id

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/arwy.arwy_4_23

Rights and Permissions

How to cite this article:
Solanki SL. Artificial intelligence in airway management and anaesthesia. Airway 2023;6:1-3

How to cite this URL:
Solanki SL. Artificial intelligence in airway management and anaesthesia. Airway [serial online] 2023 [cited 2023 Jun 7];6:1-3. Available from: https://www.arwy.org/text.asp?2023/6/1/1/374366

Artificial intelligence (AI) is based on reinforced learning and revolves around the usage of algorithms. AI works like the brain's complex neural network and uses artificial 'neurons' that learn by comparing themselves to the desired output and 'reinforcing' useful connections, thus creating the basis of the artificial neural network approach. The healthcare field generates 'big data' that includes all records. The vast spectrum of AI and the ability of deep neural networks to analyse perioperative data are being utilised for predicting outcomes and reducing healthcare costs.[1] The involvement of AI in anaesthesia practice includes preoperative assessment and risk predictions, airway management using intubating robots, managing effective concentration of anaesthesia by target-controlled infusion, monitoring the depth of anaesthesia (DOA) and early detection of anaesthesia and surgical complications. Postoperatively, AI has shown a role in reducing surgical site infections and intraabdominal infections[2] and identifying anastomotic leaks.[1]

In this issue of 'Airway', Senthilnathan and Kundra[3] discuss the role of machine learning (ML) and AI in the field of airway management and anaesthesia practice by supervised ML (the machine predicts the known outcome, e.g., regression algorithms) and unsupervised ML (no predefined outcome). They mention that the prediction of the difficult airway (DA) by ML algorithms is one promising tool that can successfully predict DA, difficult laryngoscopy and difficult intubation, which can be achieved by incorporating all the DA predictors (physical and radiological).

Senthilnathan and Kundra[3] also highlight the importance of ML and AI in anaesthesia practice and intensive care units by a decision tree, a supervised ML technique for analysing classification-related tree and regression problems. The outcome variable is categorical in the classification tree and continuous in the regression tree.

Anaesthesia is the one field, which has the maximum scope of ML and AI. The use of high technology in anaesthesia and intensive care units makes it relatively easy to collect and analyse lots of real-time unbiased data directly from multipara monitors, infusion devices, drug-delivery systems, closed-loop anaesthesia delivery systems (CLADS)[4] and electronic medical records. Ramaswamy et al.[5] used ML to analyse the complex EEG signal after giving sedation and evaluated the level of sedation in 102 volunteers. The outputs were integrated into a complex yet defined algorithm to continuously estimate the level of sedation without bias and interobserver variability and without stimulation of the subject.

ML is done in three processes: (a) Big Data Analysis-Data Mining-Cluster Analysis: in this, the large database is analysed (big data analysis), and the patterns of and between variables observed (data mining) to make or recognise a cluster of data for further analysis (cluster analysis); (b) Analysis of Complex Data from clinical sources such as multipara monitor (haemodynamic parameters, DOA, neuromuscular monitoring, etc.), EEG, degree of sedation, etc., and (c) Model or Algorithm Description for real-time prediction or estimation of a particular outcome or event.[6]

ML has very high analytical ability, even better than many statistical and modelling tools. The main issue is the 'black box problem' of AI in healthcare, which is the inability of patients or healthcare providers to understand the outcome or event of the ML or AI which is based on unspecified mathematical functions, algorithms or models.[6]

Anaesthesia practices are like the aviation sector, with considerable risk from induction to maintenance to recovery. For a safe perioperative period, Gambus and Jaramillo[6] described 'reactive, proactive and predictive' model, where the 'proactive approach' prevents an adverse outcome or event by preparing the required things beforehand like checking the anaesthesia machine, drug preparation, assessment of the patient, etc. The 'reactive approach' is reacting to certain critical events. The 'predictive approach' provides information and alerts the physician about the anticipated adverse outcomes or critical events to take preemptive preventive action. One example of the predictive approach is using Hypotension Prediction Index in the perioperative period, which alerts the physician about the possibility of hypotension.[7]

The predictive approach improves the proactive practice using a model to assess the risk of adverse events or outcomes (risk of DA, etc). The better predictive model is one that can have an effect on the reactive approach by providing information about the possibility of an adverse event or outcome before its occurrence (HPI model for 'risk of hypotension', the bispectral index for 'risk of deep sedation').

Brown et al.[8] proposed an AI system that identifies the ETT, trachea and carina using a 'semantically embedded neural network' that combines deep convolutional neural networks. It provides two forms of decision support: ETT detection assistance and position check alerts. It focuses on supporting the intensive care unit physician at the point of care. Based on the relationships defined in the semantic network, segmentation proceeds in a hierarchical fashion, with the segmentation of the trachea spatially guiding the convolutional neural networks detecting the carina and ETT. Spatial relationships are used directly to define the ETT tip and safe zone regions.

Mlodzinski et al.[9] surveyed the use of ML/AI for a novel intubation prediction tool amongst critical care physicians (through email) and critical care providers and non-providers by one open social media survey. They found that many critical care providers and non-providers have positive perceptions of ML/AI-based tools. Critical care providers were interested in tools to predict the need for intubation.

The use of robots in anaesthesia practice is not very new. The first robotic intubation was done more than a decade ago using the da Vinci surgical system (multiple arms). The Kepler intubation system is a single robotic arm controlled by a joystick for remote intubation.[10] The Magellan system has been used for regional techniques with a block needle mounted on a robotic arm. It was used with a good success rate and less interoperator variability.[11] Robotic endoscope-automated laryngeal imaging for tracheal intubation is a video-endoscopic stylet that guides endotracheal intubation.[12]

There are other uses of ML/AI in airway management, such as VivaSight single-lumen ETT[13] and VivaSight double-lumen tube.[14]

In conclusion, the use of ML algorithms and AI in anaesthesia practice is increasing, and the future is with AI. It opens the door to improved close monitoring, better drug and anaesthetic delivery, preemptive approach to maintain haemodynamics and less airway catastrophe by better assessment tools and intubation assistance.

  References Top

Solanki SL, Pandrowala S, Nayak A, Bhandare M, Ambulkar RP, Shrikhande SV. Artificial intelligence in perioperative management of major gastrointestinal surgeries. World J Gastroenterol 2021;27:2758-70.  Back to cited text no. 1
da Silva DA, Ten Caten CS, Dos Santos RP, Fogliatto FS, Hsuan J. Predicting the occurrence of surgical site infections using text mining and machine learning. PLoS One 2019;14:e0226272.  Back to cited text no. 2
Senthilnathan M, Kundra P. Predictive machine learning algorithms in anticipating problems with airway management. Airway 2023;6:4-9.  Back to cited text no. 3
  [Full text]  
Puri GD, Mathew PJ, Biswas I, Dutta A, Sood J, Gombar S, et al. A multicenter evaluation of a closed-loop anesthesia delivery system: A randomized controlled trial. Anesth Analg 2016;122:106-14.  Back to cited text no. 4
Ramaswamy SM, Kuizenga MH, Weerink MA, Vereecke HE, Struys MM, Nagaraj SB. Novel drug-independent sedation level estimation based on machine learning of quantitative frontal electroencephalogram features in healthy volunteers. Br J Anaesth 2019;123:479-87.  Back to cited text no. 5
Gambus PL, Jaramillo S. Machine learning in anaesthesia: Reactive, proactive predictive! Br J Anaesth 2019;123:401-3.  Back to cited text no. 6
Gangakhedkar GR, Solanki SL, Divatia JV. The use of hypotension prediction index in cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC). Indian J Anaesth 2022;66:294-8.  Back to cited text no. 7
  [Full text]  
Brown MS, Wong KP, Shrestha L, Wahi-Anwar M, Daly M, Foster G, et al. Automated endotracheal tube placement check using semantically embedded deep neural networks. Acad Radiol 2023;30:412-20.  Back to cited text no. 8
Mlodzinski E, Wardi G, Viglione C, Nemati S, Crotty Alexander L, Malhotra A. Assessing barriers to implementation of machine learning and artificial intelligence-based tools in critical care: Web-based survey study. JMIR Perioper Med 2023;6:e41056.  Back to cited text no. 9
Hemmerling TM, Wehbe M, Zaouter C, Taddei R, Morse J. The kepler intubation system. Anesth Analg 2012;114:590-4.  Back to cited text no. 10
Morse J, Terrasini N, Wehbe M, Philippona C, Zaouter C, Cyr S, et al. Comparison of success rates, learning curves, and inter-subject performance variability of robot-assisted and manual ultrasound-guided nerve block needle guidance in simulation. Br J Anaesth 2014;112:1092-7.  Back to cited text no. 11
Biro P, Hofmann P, Gage D, Boehler Q, Chautems C, Braun J, et al. Automated tracheal intubation in an airway manikin using a robotic endoscope: A proof of concept study. Anaesthesia 2020;75:881-6.  Back to cited text no. 12
Zang Q, Cui H, Guo X, Lu Y, Zou Z, Liu H. Clinical value of video-assisted single-lumen endotracheal intubation and application of artificial intelligence in it. Am J Transl Res 2022;14:7643-52.  Back to cited text no. 13
Saracoglu A, Saracoglu KT. VivaSight: A new era in the evolution of tracheal tubes. J Clin Anesth 2016;33:442-9.  Back to cited text no. 14


Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

  In this article

 Article Access Statistics
    PDF Downloaded80    
    Comments [Add]    

Recommend this journal